Limitations on motion sensing imposed by retinal and central visual processing.

摘要

Sensing motion generates a cascade of activity in thousands of neurons across multiple stages of the visual system, but little is known about how processing at each stage limits the fidelity of perception. We examined the precision of motion encoding in retinal signals and the variability of behavioral speed readout to explore how the mechanisms of visual computations affect the fidelity of speed encoding. Multi-electrode retinal recordings were used to measure the activity of ∼100 parasol retinal ganglion cells (RGCs) simultaneously in isolated macaque retinas stimulated with moving bars. Human psychophysics experiments using stimuli matched for eccentricity, size, and intensity, were performed using identical moving bar stimuli. RGC population activity encoded speed with a precision of ∼1%, contrasting with a behavioral precision of ∼7%. The retinal-motion signal was conveyed in ∼10 milliseconds, comparable to the inter-spike interval. Temporal structure in retinal spike trains provided more precise speed estimates than time-varying firing rates. Correlated activity between RGCs had little effect on speed estimates. The spatial dispersion of RGC receptive fields influenced speed estimates more strongly along the axis of motion than along the orthogonal direction, as predicted by a simple model based on RGC response-time variability and optimal pooling. ON and OFF cells encoded speed with similar and statistically independent variability. Simulation of downstream speed estimation, using populations of speed-tuned units, showed that peak (winner-take-all) readout provided more precise speed estimates than centroid (vector-average) readout. Fixational eye movements increased fractional behavioral speed-estimate variability ∼1%. A moving-texture stimulus was further used to probe behavioral estimation of speed. The fidelity of speed estimation was found to improve with temporal integration at the level of motion detection and spatial pooling across motion detectors. These findings account for the fidelity of speed estimates that are accessible behaviorally.